9.3 Analyzing learners' language

　 Learners' language provides data for research into the nature of the learning process. In order to gain insights into the process, researchers have engaged themselves in the analysis of learners' language. Since the 1940s, there have been three modes of analysis --- contrastive analysis, error analysis, and the study of interlanguage, each of which is a milestone in understanding second language acquisition.

　 Contrastive analysis was conducted from the 1940s to the 1960s as an effort to understand the source of errors in learners' language. It was assumed that the greater the difference between the native language and the target language, the greater the learning problem and the potentiality of mother tongue interference. The assumption was challenged by findings of later studies. Many errors were found that had no relation with the native language. As interference makes up a small proportion of errors, the contrastive analysis hypothesis has proved less powerful in explaining why learners' language is as it is.

　 Error analysis was employed from the late 1960s as part of the methodology of the study of second language acquisition. Corder (1967, 1973) pointed out that errors have theoretical and practical significance. They are theoretically significant in that they provide feedback to psycholinguistics in constructing theories of SLA. They are of practical significance to language teachers, indicating to them the effectiveness of the teaching materials and techniques.

　 The practice of error analysis is divided into identifying, describing and explaining. Identifying errors is the first step of error analysis. To identify errors we have to compare the sentences produced by the learner with the corresponding sentences native speakers are likely to produce to express the same meanings. This is not always straightforward. Sometimes it is not clear whether a deviation is a slip of the tongue or a systematic error. Corder (1973) first distinguished errors from mistakes. Errors reflect gaps in a learner's knowledge of the target language. The learner does not know what is correct. Mistakes reflect occasional lapses in performance. The learner knows the correct form but slips due to nervousness, carelessness or tiredness. For example, many Chinese learners of English, even advanced learners, use he to refer to a female person in conversations. This is a mistake, not an error, as they know what the correct form is. Learners may correct themselves once they have the time to monitor their own speech or writing. So, accurate identification of errors as the starting point of error analysis is not as easy as assumed. It is of importance to both researchers and teachers. For researchers, it is important to collect the right data. For teachers, it is necessary to treat mistakes and errors differently in their instruction.

　 Describing errors, the second step of error analysis, is categorizing errors grammatically. Once errors are identified, they can be classified into categories. Corder (1973) proposed four major categories: omission of some required elements, e.g. “He went bus stop”; addition of some unnecessary or incorrect element, e.g. “Does he can swim?”; selection of an incorrect element, e.g. “I lost my road”; and misordering of elements, e.g. “I gave to him the book”. These categories are highly generalized. Another way to describe errors is to classify them into grammatical categories. For example, we can gather all the errors that have been identified relating to verbs and then classify them. The latter is more practicable for language teachers and provides more useful feedback to teaching.

　 Explaining errors, the final step, is the task of tracing the source of errors. This task is more psychological than linguistic in essence. In terms of sources, errors are divided into interlingual errors and intralingual errors, based on whether they are caused by L1. Interlingual errors are caused by mother tongue interference. One's knowledge of L1 contributes to learning L2. The positive role L1 plays is called transfer. The negative role is termed interference. For example, many Chinese learners of English use although and but in the same sentence. This is an instance of interference, specifically, a kind of negative transfer of the learners' syntactic knowledge. Interference also occurs in other aspects of language.

　 Intralingual errors are produced by second language learners regardless of their mother tongue. If a learner says “I eated too much”, he has overgeneralized the formation of past tense. Overgeneralization is found universal in SLA (also in L1 acquisition). Some errors are attributed to simplification (also called redundancy reduction). Many Chinese learners of English omit the third person singular -s in speech and writing. This is a typical error of simplification. The meaning of third person singular is already expressed by the subject, the marker of the predicate verb is redundant in terms of information. That is why the -s is often left out. Some errors are attributed to cross-association, confusion in memory of two forms. The word stalagmite is often confused with stalactite. Dessert (sweet food) and desert (a large area of sand) are often mixed in pronunciation.

　 Although error analysis has gained some insights into the complex process of SLA, it is not without limitations. By focusing only on errors, researchers may loose sight of the whole picture of learners' language. Engaged in error analysis, researchers study what learners are doing wrong, but not what they have done successfully. Another flaw in error analysis is that it fails to account for all the areas of L2 in which learners have difficulty. Schachter (1974) reported that Chinese and Japanese learners of English committed fewer errors in relative clauses than Spanish and Persian learners of English. It was discovered that the Chinese and Japanese students avoided producing relative clauses. This result shows that fewer errors may not necessarily prove achievement in learning a particular aspect of L2. It is equally important to determine whether the learner's use of correct forms approximates that of the native speaker. This idea gave rise to the study of interlanguage.

　 Interlanguage is the approximate language system that the learner constructs for use in communication through the target language. The term was coined by the American linguist Larry Selinker, (Selinker 1972). Corder (1971) called learners' language an idiosyncratic dialect. Both terms suggest that learners' language is between L1 and L2 and that it is a continuum along which all learners traverse. Learners construct a series of mental grammars as they gradually accumulate their knowledge of the target language.

　 From the perspective of interlanguage, errors can be seen as the evidence of learning strategies. Overgeneralization reflects learners' cognitive activity in working out the rules of L2. Omission errors suggest that learners are in some way simplifying the learning task by ignoring some grammatical morphemes that they are not yet ready to use.

　 Studies in interlanguage find that learners resort to communication strategies. In communication through the target language learners every now and then are at loss in saying what they want to say in the language due to inadequate knowledge. In this case they will naturally employ communication strategies to continue the conversation. They may avoid a particular syntactic structure. They may use a superordinate (for example, worm for silkworm) or coin a word (airball for balloon, apricot seed for almond, for instance). The choice of communication strategies reflect the learner's stage of development along the interlanguage continuum. The effect of using communication strategies on SLA will be an interesting topic in SLA research.